The group deals with the quantitative methodologies for Data Science and Artificial Intelligence with applications to economics, finance and social sciences. The tools and concepts used come from mathematical statistics, statistical physics and mechanics, time series theory, operational research, theory of random and complex networks. The interest of the group is directed both to the theoretical development of new mathematical, statistical and computational methods and to the empirical study of socio-economic systems such as interaction networks between people, between companies or more generally between agents.
Among the central themes for the group there is precisely that of distinguishing, starting from the data, a system with relevant interactions from one without, measuring the interactions and using them with predictive intent.
For machine learning the group is interested in the theoretical understanding of deep learning tools interpreted as an inverse problem of statistical mechanics with boundary conditions assigned, of the relationships between deep learning and statistical and econometric methods, both inferential and hypothesis testing. Other important topics for the group include the inference of high-dimensional models including, for example, dynamic networks and variable parameter time series models, development of clustering methods and entropy and statistical approaches for the prediction / validation of interactions and hypothesis testing on networks.
In the theoretical field, soft computing techniques are also studied for decisions in an imprecise context formalized through fuzzy logic.